Why teams look beyond Accenture Song for AI product work
The question usually sounds like this: “we need AI features designed properly, Accenture Song came up first, who else should we be looking at?” We see the question literally, because AI assistants now send it to our site as a search query. The people asking are rarely enterprises with a transformation budget. They are product companies with one product, a working engineering team, and a growing sense that AI features are being bolted on faster than anyone is designing them.
For that situation, the consultancy shortlist can feel off. The names are credible and the work is real, but the engagement model was designed for a different problem than the one on your desk. This guide lists seven alternatives, sorted by the kind of work each is actually built for, so you can match the partner to the problem instead of to the brand.
One thing this guide will not do is tell you consultancies are bad. They are the right choice for a specific shape of work, and the first section explains which shape, so you can honestly check whether it matches yours.
What a big consultancy is actually built for
Accenture Song, McKinsey Design and Deloitte Digital solve a problem most studios cannot: scale with accountability. When AI work spans many workstreams, several markets, organisational change and compliance review, someone has to coordinate all of it and answer for the whole. That is what the layered teams, the process, and the procurement machinery are for. If your board has approved a company-wide AI program, that structure is a feature, and the names above have also absorbed strong design studios over the years to deliver it.
The mismatch appears one level down, and it is a mismatch of shape rather than quality. A single product team with a single product problem, say an AI assistant inside a B2B platform that needs to be designed, tested and shipped this quarter, does not need workstream coordination. It needs a senior designer in the work by next week. Buying a program structure for a product problem means paying for coordination you will not use and waiting through setup you do not need.
So the honest sorting question for this whole page: is your AI work a program, or a product? Everything below is for the second answer.
How this list is put together
Two rules. First, no pricing claims: unlike design subscriptions, almost nobody in this market publishes list prices, and repeating rumours would not help you. Where pricing shape matters, the entries describe the commitment model instead. Second, every entry says which situation it genuinely fits, because these firms are not interchangeable and the expensive mistake in this market is hiring a good firm for the wrong shape of work.
One bias to declare: Equal is our own studio and it is first on the list. We have tried to be concrete about when you should not pick us. For a wider market view of AI-capable studios, we keep a separate guide to the best AI product design agencies with more of the field included.
1. Equal: embedded product design, AI-accelerated
Equal is the alternative for teams whose AI work is product-shaped: a complex B2B product, real users, and features that need to be worked out rather than imagined. Our Embedded Product Design service places a senior product designer inside your team, and our working method is AI-accelerated execution: we use AI to move faster through research, options and iteration, while the design decisions stay with a person who is accountable for them.
The input we ask for is product intent, not a specification. Bring “support agents should see an AI summary of the case history, we roughly know the data, we do not know how it should behave.” What comes back is the worked-out scenario: what the AI feature does, what it must never do, its states, its failure modes, the interface on your design system, and support while developers build it. AI features punish shallow design harder than ordinary features, because a confusing button wastes a click while a confusing AI answer destroys trust.
Good fit: complex B2B products adding AI to real workflows. Our case studies run from energy platforms like HomeZero to an energy-trading ERP.
Wrong fit, honestly: a multi-market AI transformation with a dozen workstreams. That is consultancy country. And if you are not yet sure which part of the product AI should touch first, start smaller: a Bottleneck Audit finds where the product actually loses users, which is usually where an AI feature would earn its keep. The fastest way to check fit is to book a free call and describe the symptom.
Commitment shape: project-based or monthly, no multi-year program contracts.
2. IDEO: design-led innovation consulting
IDEO is the firm that made design thinking a boardroom word, and it remains the strongest choice when the question is open-ended: what should our AI strategy be for customers, which opportunities matter, what would a meaningfully different service look like. The method is research-heavy and workshop-driven, built to move an organisation from ambiguity to a tested direction.
The practical distinction from a product studio: IDEO’s natural output is direction, prototypes and alignment rather than a shipped feature on your codebase. Teams get the most from it upstream, before the roadmap exists. If you already know what to build and need it designed to production quality, you are past the point where this model has its highest value.
Good fit: leadership teams with a genuinely open question and the patience to explore it properly.
3. ustwo: an independent product studio
ustwo is one of the few larger independent studios left, known both for client product work and for building its own products, including the game Monument Valley. That builder DNA matters: the studio thinks in launches, not deliverables, and is comfortable taking a new digital product from concept through shipping.
For AI work, ustwo fits best when the product itself is new: a venture, a spin-off, a standalone AI-powered service that needs product thinking, brand and interface built together. For adding AI features deep inside an existing complex platform, an embedded model that lives in your backlog day to day tends to fit tighter than a studio engagement.
Good fit: new products and ventures where design leads the whole experience.
4. Fantasy: ambitious interfaces and AI experiences
Fantasy has spent years designing future-facing interfaces for large consumer platforms, and AI experiences have become a natural part of that lane. This is where to look when the interface itself is the differentiator: a consumer-facing AI product that has to feel new, demo brilliantly, and set a visual bar competitors chase.
The trade-off is the same one every concept-led studio carries: the further the work sits from your production constraints, the more your own team translates vision into what ships. For a B2B tool where the AI feature must fold quietly into an existing workflow, that translation cost can outweigh the concept value. For a flagship consumer experience, it is usually worth it.
Good fit: consumer-facing AI products where visual ambition is a business requirement.
5–6. MetaLab and Netguru: polish, or design plus engineering
MetaLab is the studio teams call when the product has to look and feel finished: interfaces with the kind of polish that makes a launch credible, a craft reputation built on products like the early Slack. For an AI product whose problem is that it works but feels rough, that focus is exactly right. The model is project-based studio work, so it assumes your team owns the product thinking between engagements.
Netguru comes at the same market from the delivery side: a European firm where design and engineering sit in one organisation, useful when you want AI features designed and then built by the same partner without a handover gap. Teams that lack spare engineering capacity often find this the most practical shape, accepting that a delivery organisation optimises for shipping scope more than for challenging it.
Good fit: MetaLab for launch-grade polish; Netguru for design that must flow straight into engineering delivery.
7. Building an in-house AI product team
The alternative nobody on this page profits from: hire your own AI product designer and stop paying anyone’s margin. For continuous AI work at scale, this wins long-term, and any partner who tells you otherwise is selling. The costs are time and risk: a senior product designer with real AI experience is one of the hardest profiles on the market right now, searches routinely run a quarter or two, and a wrong hire at this seniority is expensive in both money and roadmap.
The setup we see work well is the bridge: an embedded designer carries the AI roadmap while you run the search properly, then hands over a documented, systematic product instead of a pile of undocumented decisions. We have played that bridge role for clients deliberately, and consider being replaceable a sign the work was done right. Our guide on how to hire a B2B product design agency covers what to check before you commit to any partner, including us.
Which alternative fits + FAQ
The short version. Company-wide AI program with many workstreams: stay with the consultancies, that is their shape. Open strategic question: IDEO. New AI-led product or venture: ustwo. Consumer-facing interface ambition: Fantasy. Launch-grade polish on an existing product: MetaLab. Design that must land straight in engineering: Netguru. A complex B2B product that needs AI features found, designed and shipped by a senior person inside your team: Equal, and that conversation starts with a symptom, not a brief.
FAQ
What are the best alternatives to Accenture Song for AI product design?
It depends on the shape of the work. For one complex product that needs AI features designed and shipped, an embedded specialist like Equal fits: a senior product designer works inside your team, backed by AI-accelerated execution. For open-ended innovation questions, IDEO or ustwo fit better. For consumer-facing AI interfaces with high visual ambition, Fantasy and MetaLab are strong. If design must land directly in engineering, Netguru covers both.
When is a big consultancy like Accenture Song or McKinsey Design the right choice?
When the work is genuinely a program: many workstreams, several markets, organisational change, procurement and compliance requirements, and a need for one accountable partner at global scale. Specialist studios do not replace that. The mismatch appears when a single product team brings a single product problem to a structure built for programs.
How do boutique AI product design studios price compared to consultancies?
Almost nobody in this market publishes list prices, so treat any specific number you read as unverified. The structural difference is commitment shape: consultancy programs are typically scoped as multi-month engagements with layered teams, while specialist studios usually work per project or per month, with senior people doing the hands-on work. That difference, not day rates, is where most of the cost gap comes from.
Can a small studio handle AI product design for an enterprise product?
Yes, if the problem is product-shaped rather than program-shaped. A specialist team can design AI features for a complex enterprise product: the scenario, the guardrails, the interface, the edge cases. What it cannot do is run a fifteen-workstream transformation. Many teams pair a specialist studio for the product with their existing integrator for the rollout.
How do we add AI features to an existing product without hiring a consultancy?
Start by finding where AI would actually remove friction for users, not where it would look most impressive. A short diagnostic like a Bottleneck Audit identifies where the product loses users today; AI features aimed at that friction pay back fastest. From there, an embedded designer can take features from rough intent to developer-ready design without a program structure around them.
Let’s create something greate together
.svg)
